The present embodiments relate to identifying parts in a photograph of an assembly. Various image processes have been proposed to recognize objects in an image. In computer vision, a visual search ranks similar images in an image database and uses indexing concepts. Metadata may be assigned to photographs. In location aware applications, three-dimensional models are used to render synthetic views to increase the coverage of the potential search space. Texture or other characteristics of identified objects may be extracted from the image. Image processing approaches may have difficulty dealing with common variations in photographs and/or may not provide sufficient information about the objects in the photograph.